Systematic Review and Meta-Analysis of Machine Learning Models for Deep Vein Thrombosis

An Liu's systematic review and meta-analysis evaluates machine learning-based prediction models for deep vein thrombosis (DVT) in hospitalized patients. The review includes 17 studies, with 10 models meta-analyzed, and was last updated on 2026-05-26. It assesses methodological quality, predictive performance, and clinical applicability of these models.

Use Cases

• Benchmarking new DVT prediction models based on reported performance metrics like AUC.
• Identifying common predictive features for DVT risk based on frequently reported predictors like D-dimer, surgery, and age.
• Assessing methodological quality and reporting standards in clinical ML studies using the PROBAST tool.
• Designing multicenter external validation studies based on identified limitations in current literature.

Strengths

• Includes a meta-analysis of 10 validated models, reporting a pooled AUC of 0.85.
• Systematically reviews 17 studies, identifying D-dimer, surgery, and age as the most frequent predictors.
• Uses established methodological frameworks (PRISMA, PROBAST) for quality assessment.

Limitations

• The primary data file is a 839.0 KB DOCX document, not a structured dataset, limiting direct machine-readability.
• Six of the reviewed studies were judged to be at high risk of bias, indicating methodological weaknesses in the field.
• Substantial heterogeneity was reported in the meta-analysis, which may limit generalizability of the pooled performance estimate.

Provenance

Source

figshare, authored by An Liu.

Collection Method

Systematic literature review and meta-analysis of studies from PubMed, Embase, Web of Science, and CINAHL.

Time Range

Literature search covers from November 1, 2015, to November 1, 2025.

Freshness

Last updated 2026-05-26 04:39:01.